The specification relates to a navigation system. In particular, the specification relates to a system and method for navigating a mobile automated system through obstacles.
It is a challenge to navigate a mobile automated system (e.g., a robot) in a dynamic environment that includes dynamic obstacles such as human crowds, a combination of static obstacles and moving obstacles, etc. Generally, only a very limited amount of information regarding the dynamic environment is available since the environment keeps changing continuously. For example, the number of moving objects (e.g., crowds of walking people) in the environment changes all the time and the paths and the destinations of the moving objects are unknown. This dynamic change in the environment presents a challenge to determine a safe and collision free path for navigating the mobile automated system to a destination while respecting social behavior norms.
Existing solutions fail to provide an effective approach for navigating a mobile automated system in such a dynamic environment having heavy crowds. Furthermore, existing solutions requires training data to learn motion patterns in the dynamic environment which makes the performance dependent on the quality of the training data. Existing solutions provides non-optimal performance when there is no training data available and/or when the quality of the training data is low.